3D UAV Navigation with Moving-Obstacle Avoidance Using Barrier Lyapunov Functions

We address the safe-navigation problem for aerial robots in the presence of mobile obstacles. Our approach relies on an original dynamic model defined in a cylindrical-coordinate space. It is assumed that the environment contains moving obstacles, that are encoded as state constraints so that they are embedded in the control design: the controller is constructed so as to generate a force field which, in turn, is derived from a potential with negative gradient in the vicinity of stable equilibria and positive gradient in the vicinity of obstacles. In particular, we combine the so-called Barrier Lyapunov Functions (BLF) method with the backstepping technique to obtain a smooth time-invariant controller. It is guaranteed that the robot reaches its destination from any initial condition in the valid workspace (that is, the environment stripped of the obstacles' safety neighborhoods) while avoiding collisions. Furthermore, the performance of our control approach is illustrated via simulations and experiments on a quadrotor benchmark.